Introduction

I.1. Underwater challenges

"On peut braver les lois humaines, mais non résister aux lois naturelles."

We may brave human laws, but we cannot resist natural ones.

Twenty Thousand Leagues Under the Sea, Jules Verne

I.1.1. In the vastness of the unknown

95%. This striking figure, stated1 by the American National Oceanic and Atmospheric Administration (NOAA), tells us how little we know about oceans: about 95% of this underwater realm remains unseen by human eyes. Yet, it covers two-thirds of the Earth's surface. It is even said that we know the Moon's surface better than our oceans' depths. Nevertheless, marine technologies have changed dramatically over the last 100 years, discovering ways to explore bodies of water that previously would have been unimaginable.

We could say that the underwater exploration started with the Challenger Expedition (1872, Figure I.1) by probing the depths from the surface with lead lines. The Challenger Deep, which is the deepest known point on Earth2, was discovered during this expedition. Yet, it was not until the start of the 1960s that this spot was visited by humans, during the dive of the manned submersible Trieste (Figure I.2). Ever since, the place has been reached by very few expeditions, mainly unmanned descents.

Figure I.1. The HMS Challenger, a British corvette that took part in the first global marine research expedition: the Challenger Expedition, 1872-1876. Painting by William Frederick Mitchell. For a color version of the figures in this chapter see www.iste.co.uk/rohou/robot.zip

Figure I.2. Trieste is a Swiss-designed and Italian-built deep-diving research bathyscaphe. It was able to reach any point of the Earth's abysses, such as the Mariana Trench in 1960. Photo: U.S. Naval Historical Center

The dive of the Trieste revealed the capacity to build vehicles that are able to resist colossal pressures. However, the cost of this endeavor is huge when compared to the range of the explored area: only a few square meters around the submersible. If exploration techniques have evolved considerably over the years, the ratio of exploration/cost or exploration/time remains a major impediment to the discovery of our oceans.

I.1.2. Hostile environments

Withstanding the high pressures of the column water, corrosive salinity, unpredictable currents, etc. is one thing; perceiving the environment is another. Figure I.3 provides an example of poor visibility that can be encountered under the surface. Strong opacities in shallow waters, or lack of light in the deepest ones, make it difficult to gather information from cameras. Other conventional means of exploration or communication suffer from strong attenuations of their electromagnetic waves through the water column.

Underwater acoustics

Underwater acoustics is the only technology left with sufficient performances to increase the range of visibility. A telling experiment is the Heard Island test performed in 1991 (Munk et al. 1994), which was planned in order to test the emission of an artificial acoustic signal in the world's oceans. A special phase-modulated signal of 57 Hz, emitted from an island located in the southern Indian Ocean, was received by 16 sites around the world, some of them were based on the two coasts of North America. This experiment demonstrated that great distances can be reached by acoustics.

Considering an estimation of the sound celerity profile along the propagation, an acoustic wave is even well suited to perceive distances between the emitter and any obstacle in the environment. In practice, ranges of a few dozen meters are affordable to maintain precision at a reasonable energy cost. However, we should note that an acoustic signal rarely propagates in a straight line. This has an impact on estimation of distances and may even generate blind zones3. Underwater acoustics nonetheless remains the most suited approach for wide explorations, but the related solutions are far from being straightforward.

Figure I.3. In the shallow waters of La Spezia (Italy) during the SAUC-E competitions in the NATO Centre for Maritime Research and Experimentation (CMRE, formerly NURC), 2013-2014. These images were taken by the ENSTA Bretagne's autonomous robot Vici. Designing algorithms to automatically analyze these observations remains a challenging task

A needle in a haystack

The work presented in this book started on the very same day that the underwater search began for the lost MH370 aircraft operated by Malaysia Airlines, which presumably disappeared in the southern Indian Ocean in 2014. Despite a tremendous deployment of maritime means, making this multinational search effort the largest and most expensive in aviation history, the aircraft remains unfound. From October 2014 to January 2017, an overall survey of 120,000 km2 of the seafloor was performed with unsuccessful results. Given the vast areas involved, this search sadly reveals the difficulty we still have in exploring the extent of the seabed.

Figure I.4. Extract from the bathymetric survey conducted during the search for MH370 aircraft off the west coast of Australia. Gray areas represent the bathymetry that was indirectly estimated using satellite-derived gravity data. In contrast, colored data were acquired by marine means, highlighting the need to undertake surveys in situ for higher precisions. ©Copyright 2014, Commonwealth of Australia

The unfruitful research nonetheless improved the knowledge we had on this part of the oceans, providing a level of details that had rarely been reached in the deep environment (Picard et al. 2017). Figure I.4 shows a comparison between the previous mapping of the seabed, which had an average spatial resolution of about 5 km2, and the new digital elevation model (DEM) obtained with a resolution of less than 0.01 km2. During the search, the vessels equipped with acoustic means, such as side-scan sonars or multibeam echosounders, were not able to scan the entire extent of the search area. Indeed, the seabed parts with the most complex and challenging topography could only be reached by autonomous underwater vehicles (AUVs), equipped with similar technology and specifically designed for high-resolution survey operations in remote deep water locations. These vehicles lend a helping robotic hand in such exploration efforts.

I.1.3. Autonomous underwater vehicles

Owing to the difficulties posed by complex environments and vast areas that are still uncovered, the use of autonomous vehicles appears to be a durable solution to face these challenges and push the boundaries of the knowledge of the oceans. Indeed, even with efficient methods such as underwater acoustics, the footprint of marine sensors is still modest in view of the extent of what has to be explored. Multiplying the number of vessels equipped with sensors is expensive due to the involvement of crew. In addition, surface vehicles are not sufficient to provide the details of deep waters. Marine robots (Creuze 2014) are an attractive alternative to increase the exploration means at a reasonable cost.

Furthermore, global supervision of an underwater robot performing an exploration task is rarely affordable due to the opacities of the environment mentioned previously. The low rate of underwater communications and the latency during the propagation of messages require the robot to possess a full degree of autonomy. For these reasons, new marine robots are designed to make unsupervised decisions in order to achieve a given task. They can be involved in several marine applications such as hydrography, oceanography, climate change monitoring, military operations in mine hunting (Toumelin and Lemaire 2001), wreck searches (L'Hour and Creuze 2016), etc.

Because they sail underwater without receiving orders from the surface, AUVs need to sense their environment and act accordingly; thus, they are equipped with sensors such as sonars or cameras. In addition, they estimate their own position by themselves (Leonard et al. 1998), which is always a complicated task in the underwater world. The localization problem will be presented in section I.2, which is the main motivation of this book. The contributions of this work will be presented through actual experiments involving two AUVs4, Redermor and Daurade, which are introduced below.

The Redermor AUV

The Redermor5 AUV, shown in Figure I.5, was an experimental robot designed during the Franco-British collaborative project Remote Mine Hunting System. Built during the 1990s at DGA Techniques Navales Brest (formerly GESMA), it served as a platform for several studies (Quidu et al. 2007). The main characteristics of the vehicle are summarized in Table I.1, (Toumelin and Lemaire 2001).

Figure I.5. The Redermor AUV before a sea trial. The thrusters' layout allows it to circumnavigate a point such as a mine to be identified, its front-looking sonar providing different viewing angles of the target. Photo: DGA-TN...